INDEX

Explanations

No Explanations Found

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.hook_resid_pre

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

24,576

Data Type

torch.float32

Hook Point

blocks.0.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

ciating

-0.73

 granddaughter

-0.68

 Lisbon

-0.67

zhen

-0.66

IPM

-0.66

 daughter

-0.63

 firewall

-0.61

daughter

-0.61

rolet

-0.61

agate

-0.60

POSITIVE LOGITS

Ax

0.71

 Parade

0.69

âķĲâķĲ

0.69

estones

0.66

aurus

0.64

abba

0.62

èĢħ

0.62

Tes

0.61

Pear

0.61

kas

0.61

Activations Density 0.000%

No Known Activations

This feature has no known activations.